Before HIGGINBOTHAM, SMITH, and OLDHAM, Circuit Judges. JERRY E. SMITH, Circuit Judge: The United States is a few days from the November 3, 2020, General Election. Texas officials are preparing for a dramatic increase of mail-in voting, driven by a global pandemic.[1] One of their many pressing concerns is to ensure the integrity of the ballot by adhering to the state’s election-security procedures. And the importance of electoral vigilance rises with the increase in the number of mail-in ballots, a form of voting in which “the potential and reality of fraud is much greater . . . than with in-person voting.” Veasey v. Abbott, 830 F.3d 216, 239 (5th Cir. 2016) (en banc). “Absentee ballots remain the largest source of potential voter fraud “[2] In a well-intentioned but sweeping order issued less than two months before the election,[3] however, the district court minimizes Texas’s interest in preserving the integrity of its elections and takes it upon itself to rewrite the Legislature’s mail-in ballot signature-verification and voter-notification procedures. At a time when the need to ensure election security is at its zenith, the district court orders that, if the Secretary of State does not adopt its preferred procedures, election officials must stop altogether rejecting ballots with mismatched signatures. Because Texas’s strong interest in safeguarding the integrity of its elections from voter fraud far outweighs any burden the state’s voting procedures place on the right to vote, we stay the injunction pending appeal. I. Though it is not constitutionally required to do so,[4] Texas offers qualifying citizens the option to vote by mail. See TEX. ELEC. CODE §§ 82.001–004.[5] Specifically, the state extends the privilege to over-65 voters, the disabled, and those either in jail or otherwise absent from their county on election day. Id. To further its compelling interest in safeguarding the integrity of the election process, Texas conditions the vote-by-mail privilege on compliance with various safeguards. One of those, at issue here, is signature verification. To vote by mail, a voter must first apply for a mail-in ballot. § 84.001(a). The applicant must sign, § 84.001(b), and timely submit the application by mail to the early voting clerk, § 84.001(d). A witness may sign the application if the applicant cannot sign because of physical disability or illiteracy. § 1.011(a). Once a voter applies and is deemed eligible to vote by mail, the Early Voting Clerk must provide the balloting materials to the voter by mail. §§ 86.001(b), 86.003(a). Included in those materials are the ballot, ballot envelope, and carrier envelope. § 86.002(a). After receiving those materials, a voter who wishes to cast a ballot must fill out the ballot, seal the ballot envelope, place it in the carrier envelope, § 86.005(c), and timely return it, § 86.007. Additionally, the voter must sign[6] the certificate on the carrier envelope, § 86.005(c), which “certif[ies] that the enclosed ballot expresses [the voter's] wishes independent of any dictation or undue persuasion by any person,” § 86.013(c). The Early Voting Ballot Board (“EVBB”) is responsible for processing the results of early voting. § 87.001.[7] The Early Voting Clerk may appoint, in addition to the EVBB, a Signature Verification Committee (“SVC”). § 87.027(a).[8] Upon receipt of the mail-in ballots, the Early Voting Clerk must timely deliver the ballots to the SVC or, if no SVC is appointed, to the EVBB. §§ 87.027(h), 87.021(2), 87.022–024. If no SVC is appointed, the EVBB receives and reviews each ballot to determine whether to accept it. § 87.041(a). Relevant here, the EVBB may accept a ballot “only if . . . neither the voter’s signature on the ballot application nor the signature on the carrier envelope certificate is determined to have been executed by a person other than the voter, unless signed by a witness . . . .” § 87.041(b)(2). In making that determination, the EVBB compares the two signatures and “may also compare the signatures with any two or more signatures of the voter made within the preceding six years and on file with the county clerk or voter registrar.” § 87.041(e). If the EVBB determines that a ballot is not acceptable— as a result of either the signature-verification procedure or another of § 87.041(b)’s requirements—the ballot is rejected, and the vote is not counted. §§ 87.041(d), 87.043(c). If the Early Voting Clerk appoints an SVC, the committee receives the ballots and makes the signature-verification determination before delivering the ballots to the EVBB. § 87.027(h)–(i). The SVC follows a similar, though slightly more robust, procedure for verifying signatures than does the EVBB. Compare § 87.027(i) with § 87.041(b)(2). The Code instructs the SVC to compare the two signatures on the ballot application and the carrier envelope certificate “to determine whether the signatures are those of the voter.” § 87.027(i). It also permits the SVC to aid its determination by comparing the two signatures with “any two or more signatures of the voter made within the preceding six years and on file with the county clerk or voter registrar . . . .” Id. A determination that the signatures do not belong to the voter “must be made by a majority vote of the committee’s membership.” Id.[9] Once the SVC has made its signature-verification determinations, the committee’s chair delivers the ballots to the EVBB. Id. The EVBB follows the same procedures it otherwise would, except that it is bound by the SVC’s determination that the signatures belong to the voter. § 87.027(j). Conversely, if the EVBB believes that the SVC erroneously determined that the ballot failed the signature-verification procedure, it may reverse that determination by a majority vote and accept the ballot. Id. Thus, if either body determines that the signatures belong to the voter, that determination is final, and the ballot may not be rejected on that basis. If the EVBB rejects a ballot, it must note the reason on the carrier envelope. § 87.043(d). When its review is complete, the EVBB places the rejected ballots into an envelope or envelopes, records the number of rejected ballots in the envelope, and seals it. § 87.043(a). The EVBB then delivers the rejected ballots to the general custodian of election records. § 87.043(c). No later than ten days after the date of the election, the EVBB must provide written notice of its rejection to the voter at the address on the ballot application. § 87.0431(a). Not more than thirty days after the election, the Early Voting Clerk must notify the attorney general of the EVBB’s rejections and provide the attorney general with certified copies of the rejected voters’ carrier envelopes and corresponding ballot applications. § 87.0431(b)(3). Though a voter must receive notice that his ballot was rejected, the Code does not require an opportunity to challenge that decision.[10] II. The plaintiffs challenged Texas’s absentee-ballot system in August 2019, suing the Secretary of State, Ruth Hughs; the Brazos County Elections Administrator, Trudy Hancock; and the City of McAllen’s Secretary, Perla Lara. The plaintiffs—a group comprised of two persons who had absentee ballots rejected in previous elections and organizations involved in voter registration, education, outreach, and support—raised several claims. They maintain that Texas’s signature-comparison and voter-notification procedures violate (1) the Due Process Clause of the Fourteenth Amendment, (2) the Equal Protection Clause of the Fourteenth Amendment, (3) the Americans with Disabilities Act, and (4) the Rehabilitation Act of 1973. The district court denied defendants’ motions to dismiss, and the parties conducted discovery, which lasted until May 2020, after which both sides moved for summary judgment. In August 2020, the district court requested supplemental briefing regarding what relief it should provide if it found for the plaintiffs on the merits. Describing their proposal as a “narrowly tailored remedy,” the plaintiffs asked for an injunction requiring election officials to take various rapid affirmative steps to provide notice to voters whose ballots have been rejected, to loosen absentee voter-identification requirements, and to implement an elaborate and expedited process for challenges by voters with rejected ballots. If the court found that remedy to be “impossible, impractical, or overly burdensome,” it instead should enjoin officials from engaging in a signature-comparison process at all. The Secretary took several actions over recent months to facilitate the ability of qualifying voters to vote by mail. She provided guidance to local election officials, recommending that they notify voters of rejected ballots as quickly as possible. She reminded election officials of how early they may convene EVBBs. She also alerted local election officials that they may examine not only the signatures on a voter’s application and carrier envelope, but also other signatures on file and made within the last six years. She published a letter providing mail-in voters with guidance on how properly to complete and send their ballots and giving notice of the signature-comparison process. The district court granted the plaintiffs’ summary judgment motion in part. In its detailed and lengthy memorandum opinion and order, the court “focus[ed] its analysis only on certain Plaintiffs’ claims against” Secretary Hughs, addressing only the due process and equal protection claims of Weisfeld and the Coalition of Texans with Disabilities. Richardson, 2020 WL 5367216, at *5. The district court issued an injunction adopting many of the plaintiffs’ proposed changes to Texas’s election procedures. See id. at *38–39. The injunction contained three main provisions pertaining to the 2020 election. Id. at *37–39. First, the court required the Secretary to issue an advisory, within ten days, notifying local election officials of the injunction. Id. at *38. The notification must inform them that rejecting ballots because of mismatching signatures is unconstitutional unless the officials take actions that go beyond those required by state law. Id. Second, the district court gave the Secretary a menu of actions that she must take. The Secretary must either issue an advisory to local election officials requiring them to follow the court’s newly devised signature-verification and voter-notification procedures, or else promulgate an advisory requiring that officials cease rejecting ballots with mismatched signatures altogether. See id. Third, the court mandated that the Secretary take action against any election officials who fail to comply with the district court’s newly minted procedures. Id. at *39. Deeming those dictates “appropriate for the November 2020 elections,” the court stated that it would hold a hearing after the election to consider imposing additional long-term election procedures. Id. at *45. On September 9, the Secretary filed a notice of appeal and a motion requesting the district court to stay its order pending appeal. The district court denied a stay on September 10. On September 11, the Secretary filed in this court an emergency motion for a stay pending appeal. On September 11, this panel granted a temporary administrative stay in order to consider the motion. III. “A stay is not a matter of right, even if irreparable injury might otherwise result.” Tex. Democratic Party v. Abbott (“TDP-I”), 961 F.3d 389, 397 (5th Cir. 2020) (Smith, J.) (quoting Nken v. Holder, 556 U.S. 418, 433 (2009)). “Whether to grant a stay is committed to our discretion.” Id. (citing Thomas v. Bryant, 919 F.3d 298, 303 (5th Cir. 2019)). We assess “(1) whether the stay applicant has made a strong showing that he is likely to succeed on the merits; (2) whether the applicant will be irreparably injured absent a stay; (3) whether issuance of the stay will substantially injure the other parties interested in the proceeding; and (4) where the public interest lies.” Nken, 556 U.S. at 426. “The first two factors are the most critical.” Valentine v. Collier, 956 F.3d 797, 801 (5th Cir. 2020) (per curiam). “The proponent of a stay bears the burden of establishing its need.” Clinton v. Jones, 520 U.S. 681, 708 (1997). The Secretary is likely to succeed on the merits. At the very least, she is likely to show that the district court erred in its analysis of plaintiffs’ claims. As we recently noted in two election-related opinions ruling on motions for stays pending appeal, because the Secretary is likely to succeed on one ground, we need not address the others.[11] We therefore express no opinion on the Secretary’s arguments concerning standing or whether sovereign immunity bars the present suit against her. We do, however, examine whether the district court’s remedy is barred by sovereign immunity. A. The Secretary contends that she is likely to succeed in showing that Texas’s signature-verification procedures are constitutional. In particular, she asserts that (1) those procedures do not implicate the plaintiffs’ due process rights, (2) the Anderson/Burdick framework—as distinguished from Mathews v. Eldridge, 424 U.S. 319, 335 (1976)—provides the appropriate test for the due process claims, and (3) the signature-verification procedures withstand scrutiny under Anderson/Burdick. The Secretary will likely prevail on each point. 1. We must first determine whether the plaintiffs have alleged any cognizable interests that warrant due process analysis.[12] They have not. The plaintiffs bring procedural due process claims,[13] which require two inquiries: (1) “whether there exists a liberty or property interest which has been interfered with by the State” and (2) “whether the procedures attendant upon that deprivation were constitutionally sufficient.” Kentucky Dep’t of Corr. v. Thompson, 490 U.S. 454, 460 (1989). Because the plaintiffs “invoke [the Due Process Clause's] procedural protection,” they had the burden in the district court of establishing a cognizable liberty or property interest. Wilkinson v. Austin, 545 U.S. 209, 221 (2005).[14] The Fourteenth Amendment says that states may not “deprive any person of life, liberty, or property, without due process of law.” In its conscientious 103-page order, the district court didn’t cite the Fourteenth Amendment—the sole constitutional provision it purported to interpret on the merits—even once. It’s no surprise, then, that the court also failed to identify the category of interest—life, liberty, or property—at stake in the right to vote. The plaintiffs’ brief is similarly silent. And this court has never squarely addressed the issue.[15] It is important, however, to identify a cognizable interest under the Due Process Clause, because we often dismiss due process claims where plaintiffs fail to identify a cognizable interest[16] and because “[t]he types of interests . . . for Fourteenth Amendment purposes are not unlimited.” Thompson, 490 U.S. at 460. No protection of life is raised, so we examine property and liberty interests. a. Property interests “are not created by the Constitution.” Bd. Of Regents of State Colls. v. Roth, 408 U.S. 564, 577 (1972). Instead, they “are created and their dimensions are defined by existing rules or understandings that stem from an independent source such as state law.” Id. For instance, courts sometimes consider welfare payments and continued employment to be property interests. Id. at 578. We have found no court that has held that the right to vote—much less the alleged right to vote by mail—is a property interest.[17] Neither the plaintiffs nor the district court expressly asserts that the right to vote is a property interest.[18] In fact, the complaint omits the word “property” when quoting the Fourteenth Amendment. Given the absence of argument or precedent on point, the Secretary is likely to show that plaintiffs alleged no cognizable property interest. b. Several district courts have concluded that the right to vote is a liberty interest. See, e.g., Raetzel, 762 F. Supp. at 1357. Liberty interests arise from either “the Constitution itself, by reason of guarantees implicit in the word liberty,” or from “an expectation or interest created by state laws or policies.” Wilkinson, 545 U.S. at 221 (internal quotation marks omitted). Liberty interests that arise from the Constitution extend beyond “freedom from bodily restraint.” Roth, 408 U.S. at 572. They also include the right to contract, to engage in “the common occupations of life,” to gain “useful knowledge,” to marry and establish a home to bring up children, to worship God, and to enjoy “those privileges long recognized . . . as essential to the orderly pursuit of happiness of free men.” Id. On the other hand, state-created liberty interests are “generally limited to freedom from restraint . . . .” Sandin v. Conner, 515 U.S. 472, 484 (1995).[19] This is “a narrow category of state-created liberty interests.” Jordan v. Fisher, 823 F.3d 805, 810 (5th Cir. 2016). The plaintiffs cite no circuit precedent suggesting that state-created liberty interests exist outside the context of bodily confinement. There are two problems with describing the right to vote as a liberty interest. First, the district court styled it as a state-created interest, concluding that, because “Texas has created a mail-in ballot regime,” it must now provide “due process protections.” Richardson, 2020 WL 5367216, at *21.[20] But precedent demonstrates that state-created liberty interests are limited to particular sorts of freedom from restraint. Sandin, 515 U.S. at 484. And the plaintiffs cite no binding authority indicating that state-created liberty interests exist outside the context of bodily confinement. Thus, the Secretary is likely to show that voting does not implicate any state-created liberty interest under the Due Process Clause. Second, setting aside the district court’s treatment of the right at stake, it might seem intuitive, as the plaintiffs suggest, that the right to vote is a liberty interest that arises from the Constitution. After all, the right to vote is a fundamental constitutional right. McDonald, 394 U.S. at 807. But that helps the plaintiffs with their equal protection claim, not their procedural due process claim.[21] For procedural due process, the question is not whether the plaintiffs assert a fundamental right, but instead whether the right they assert is a liberty interest. Besides describing the right to vote as fundamental, the plaintiffs have not explained what there is about the right to vote that makes it a liberty interest. The right to vote does not immediately resemble the rights described in Roth, 408 U.S. at 572. The plaintiffs cite no circuit or Supreme Court precedent extending the label of “liberty interest” to the right to vote. The Sixth Circuit, the only circuit to squarely address this issue,[22] held that the right to vote does not constitute a liberty interest.[23] Though the plaintiffs will likely run into trouble in establishing that the right to vote is a liberty interest, they will have even greater difficulty showing that an alleged right to vote by mail constitutes a liberty interest. In the context of an absentee ballot statutory scheme, “[i]t is thus not the right to vote that is at stake here but a claimed right to receive absentee ballots.” McDonald, 394 U.S. at 807. It would “stretch[] the concept too far to suggest that a person is deprived of liberty” when the Court has said that he has no right to the object of his alleged liberty interest. Roth, 408 U.S. at 572 (cleaned up). Given the failure of the plaintiffs and the district court to assert that voting—or, for that matter, voting by mail—constitutes a liberty interest, along with the absence of circuit precedent supporting that position, the Secretary is likely to prevail in showing that the plaintiffs’ motion for summary judgment on their due process claim should have been denied. c. Finally, we reject the district court’s reasoning regarding any statecreated liberty interest. The court concluded that because “Texas has created a mail-in ballot regime . . . the State must provide those voters with constitutionally-sufficient due process protections before rejecting their ballots.” Richardson, 2020 WL 5367216, at *21. That notion originated in Raetzel, in which the District of Arizona acknowledged that absentee voting “is a privilege and a convenience,” and yet concluded—without citation— “[y]et, such a privilege is deserving of due process.” Raetzel, 762 F. Supp. at 1358. In its defense, Raetzel‘s reasoning resembles the principle animating Goss v. Lopez, 419 U.S. 565 (1975). Goss concluded that, “[h]aving chosen to extend the right to an education to people of appellees’ class generally, Ohio may not withdraw that right on grounds of misconduct, absent fundamentally fair procedures . . . .” Goss, 419 U.S. at 574. Although several district courts have regurgitated Raetzel‘s reasoning,[24] the plaintiffs and the district court point to no circuit court that has embraced it. And properly so. There is a problem with grafting Goss‘s reasoning onto the voting context: Goss found two cognizable due process interests, namely a “property interest in educational benefits” and a “liberty interest in reputation.” Goss, 419 U.S. at 576. In context, Goss‘s language about the state’s “[h]aving chosen to extend” benefits and being thus bound by due process came from its analysis of a “protected property interest.” Id. at 579 (emphasis added). Raetzel, however, concluded that “the right to vote is a ‘liberty‘ interest.” Raetzel, 762 F. Supp. at 1357 (emphasis added). Thus, Raetzel grafted the Supreme Court’s reasoning concerning property interests onto a claimed liberty interest without providing any authority justifying that extension. We decline to adopt Raetzel‘s extrapolation of Supreme Court precedent. The Secretary is likely to show that the plaintiffs have alleged no cognizable liberty or property interest that could serve to make out a procedural due process claim. The Secretary is therefore likely to succeed in the dismissal of plaintiffs’ due process claims. 2. Even supposing that voting is a protected liberty or property interest, the Secretary is likely to show that the district court used the wrong test for the due process claim. The court applied Eldridge, 424 U.S. at 335, which provides the “general[]” test for determining what process is due.[25] On the other hand, Anderson v. Celebrezze, 460 U.S. 780 (1983), and Burdick v. Takushi, 504 U.S. 428 (1992) announce a test to address “[c]onstitutional challenges to specific provisions of a State’s election laws” under “the First and Fourteenth Amendments.” Anderson, 460 U.S. at 789.[26] Neither Anderson nor Burdick, however, dealt with procedural due process claims, and both instead based their approach on the “fundamental rights strand of equal protection analysis.” Id. at 787 n.7 (cleaned up). For several reasons, the Anderson/Burdick framework provides the appropriate test for the plaintiffs’ due process claims. First, because the plaintiffs challenge Texas’s election laws under the Due Process Clause of the Fourteenth Amendment, Richardson, 2020 WL 5367216, at *19, we must use the test that the Supreme Court prescribed for “[c]onstitutional challenges to specific provisions of a State’s election laws” under “the First and Fourteenth Amendments,” Anderson, 460 U.S. at 789. As several Justices have noted, “[t]o evaluate a law respecting the right to vote—whether it governs voter qualifications, candidate selection, or the voting process—we use the approach set out in Burdick v. Takushi.” Crawford, 553 U.S. at 204 (Scalia, J., concurring) (emphasis added). The district court concluded otherwise only by relying on its own word associations—with abstract concepts such as “procedures” and “procedural safeguards,” Richardson, 2020 WL 5367216, at *20—and ignoring the Supreme Court’s command that lower courts “considering a [Fourteenth Amendment] challenge to a state election law must” apply the Anderson/Burdick framework, Burdick, 504 U.S. at 434 (emphasis added). Second, our sister circuits—some of which neglected to examine whether voting constitutes a cognizable liberty or property interest—apply Anderson/Burdick to all Fourteenth Amendment challenges to election laws.[27] Although several district courts have applied Eldridge to due process challenges of signature-comparison procedures, none of those courts provided reasoning for its selection of the Eldridge test.[28] Moreover, two of those opinions cite Burdick in their due process analyses,[29] and one—though still applying Eldridge—even agreed that Anderson/Burdick applies to “all First and Fourteenth Amendment challenges to state election laws.” Frederick, 2020 WL 4882696, at *16 (emphasis added).[30] Third, even if, arguendo, we had carte blanche to decide which test applies, the Anderson/Burdick approach is better suited to the context of elecion laws than is the more general Eldridge test. “There must be a substantial regulation of elections if they are to be fair and honest and if some sort of order, rather than chaos, is to accompany the democratic processes.” Storer v. Brown, 415 U.S. 724, 730 (1974). And “it is the state legislature—not . . . federal judges—that is authorized to establish the rules that govern” elections.[31] The flaw in using Eldridge is that election laws, by nature, “inevitably affect[] . . . the individual’s right to vote.” Anderson, 460 U.S. at 788. Under Eldridge, however, courts may accord the private interest at stake, namely the right to vote, “significant weight.” Richardson, 2020 WL 5367216, at *21. Therefore, the Eldridge test would inevitably result in courts’ “weigh[ing] the pros and cons of various balloting systems,” Weber, 347 F.3d at 1107, thereby “t[ying] the hands of States seeking to assure that elections are operated equitably and efficiently,” Burdick, 504 U.S. at 433. Unlike Eldridge, the Anderson/Burdick approach recognizes that “the state’s important regulatory interests are generally sufficient to justify reasonable, nondiscriminatory restrictions.” Anderson, 460 U.S. at 788. Because Anderson/Burdick—unlike Eldridge—appropriately accounts for the state’s interest in regulating voting, it provides the appropriate test for procedural due process claims challenging election laws. By using Eldridge, the district court’s “judicial supervision of the election process . . . flout[s] the Constitution’s express commitment of the task to the States.” Crawford, 553 U.S. at 208 (Scalia, J., concurring) (citing U.S. CONST. art. I, § 4). The Secretary is thus likely to show that the district court applied the wrong test in analyzing the due process claims. 3. The Secretary contends that Texas’s signature-verification procedures withstand scrutiny under Anderson/Burdick. The parties appear to agree that Anderson/Burdick provides the appropriate framework to analyze the equal protection claims, and we have concluded that it is also the appropriate test to analyze the due process claims if the plaintiffs are able to prove a cognizable liberty or property interest. We thus analyze the equal protection and due process claims together.[32] The Anderson/Burdick rubric requires us to examine two aspects of Texas’s signature verification procedures: (1) whether the process poses a “severe” or instead a “reasonable, nondiscriminatory” restriction on the right to vote and (2) whether the state’s interest justifies the restriction. Burdick, 504 U.S. at 434 (cleaned up). Texas’s signature-verification procedures are reasonable and nondiscriminatory, and they survive scrutiny under Anderson/Burdick. a. The plaintiffs and the district court reason that Texas’s signature-verification procedures impose a severe burden on the right to vote, because “voters who have their ballots rejected due to a perceived signature mismatch are provided untimely notice of rejection and no meaningful opportunity to cure.” Richardson, 2020 WL 5367216, at *33 (emphasis omitted). Consequently, the argument goes, these voters “face complete disenfranchisement.” Id. This theory stems from two fundamental errors: It (1) mistakenly focuses only on the burden to the plaintiffs—instead of voters as a whole—and (2) neglects meaningfully to analyze binding precedent concerning what constitutes a “severe” burden on the right to vote. First, the district court concluded that Texas’s signature-verification procedures constitute “a ‘severe’ burden on certain voters’ right to vote.” Id. at *34 (emphasis added). But the severity analysis is not limited to the impact that a law has on a small number of voters. For instance, Crawford‘s three concurring Justices concluded that “our precedents refute the view that individual impacts are relevant to determining the severity of the burden” that a voting law imposes. Crawford, 553 U.S. at 205 (Scalia, J., concurring). Though Crawford‘s three-Justice plurality did not go as far as the three-Justice concurrence, it too examined the burden on “most voters.” Id. at 198. Examining burdens on a plaintiff-by-plaintiff basis “would effectively turn back decades of equal-protection jurisprudence.” Id. at 207 (Scalia, J., concurring). Specifically, the district court’s individualized assessment of burdens ignores Burdick—the very case that it purports to apply. For instance, in Burdick, 504 U.S. at 436−37, Hawaii’s ballot access laws did not constitute a severe burden on the right to vote when any burden was borne “only by those who fail to identify their candidate of choice until days before the primary.” In fact, the Burdick dissenters—whose views did not carry the day—asserted that the law’s impact on only “some individual voters” could constitute a severe burden. Id. at 448 (Kennedy, J., dissenting).[33] Thus, if we were “[t]o deem ordinary and widespread burdens like these severe” based solely on their impact on a small number of voters, we “would subject virtually every electoral regulation to strict scrutiny, hamper the ability of States to run efficient and equitable elections, and compel federal courts to rewrite state electoral codes.” Clingman v. Beaver, 544 U.S. 581, 593 (2005). Second, the plaintiffs and the district court neglect meaningfully to analyze binding precedent concerning what constitutes a “severe” burden on the right to vote. Crawford concluded that a photo-identification requirement was not a severe burden, even where “a somewhat heavier burden may be placed on a limited number of persons.” Crawford, 553 U.S. at 199. But these burdens were “neither so serious nor so frequent as to raise any question about the constitutionality” of the requirement. Id. at 197. Signature-verification requirements, like photo-ID requirements, help to ensure the veracity of a ballot by “identifying eligible voters.” Id. Signature-verification requirements are even less burdensome than photo-ID requirements, as they do not require a voter “to secure . . . or to assemble” any documentation. Id. at 199. True, some voters may have difficulty signing their names on ballots. But in Crawford, even though some voters might find it “difficult either to secure a copy of their birth certificate or to assemble the other required documentation to obtain a state-issued identification,” that difficulty did not render the photo-ID law a severe burden on the right to vote. Id. Even if some voters have trouble duplicating their signatures, that problem is “neither so serious nor so frequent as to raise any question about the constitutionality” of the signature-verification requirement. Crawford, 553 U.S. at 197–98. “[N]o citizen has a Fourteenth . . . Amendment right to be free from the usual burdens of voting.” Veasey, 830 F.3d at 316 (Jones, J., concurring) (cleaned up). And “mail-in ballot rules that merely make casting a ballot more inconvenient for some voters are not constitutionally suspect.” Tex. LULAC, 2020 WL 6023310, at *6. Moreover, Texas mitigates the burden of its signature-verification requirement in three ways. First, for those who sign ballots, the Secretary has issued notice of the signature-comparison process and guidance on how to complete a ballot properly. Richardson, 2020 WL 5367216, at *45. Second, for those who cannot sign a ballot “because of a physical disability or illiteracy,” § 1.011(a), Texas prohibits rejection of a ballot for failed signature verification if the ballot is signed by a witness, § 87.041(b)(2). Third, for those who cannot sign a ballot or who are concerned that they will be unable to provide a matching signature, Texas provides in-person voting.[34] “In Texas, in-person voting is the rule . . . [and] voting by mail is the exception.” Tex. Democratic Party v. Abbott (“TDP-II”), No. 20-50407, 2020 U.S. App. LEXIS 32503, at *1 (5th Cir. Oct. 14, 2020) (published) (citation omitted). Because Texas’s signature-verification requirement is no more burdensome on the right to vote than was the photo-ID mandate in Crawford, it does not constitute a severe burden. Instead, the signature-verification requirement is a “reasonable, nondiscriminatory restriction[]” on the right to vote. Burdick, 504 U.S. at 434. The district court, however, concluded that the signature-verification procedures constitute a severe burden because they provide “untimely notice of rejection and no meaningful opportunity to cure.” Richardson, 2020 WL 5367216, at *33. Texas could remedy that transgression, the court posited, if its mechanism for screening ballots “imposed no risk of uncorrectable rejection.” Id. at *33 n.41. But the court failed to specify how a dearth of opportunities to cure transmogrifies Texas’s signature-verification requirement into a severe burden. Similarly, the court did not cite any precedent suggesting that “no risk” of uncorrectable rejection is a constitutionally mandated standard for verifying ballots.[35] Nor could it. Indeed, the Constitution does not demand such a toothless approach to stymying voter fraud. We have found no “authority suggesting that a State must afford every voter . . . infallible ways to vote.” Tex. LULAC, 2020 WL 6023310, at *6. For instance, Crawford upheld a photo-ID law even though a voter might be unable to cast a ballot on election day because he “may lose his photo identification, may have his wallet stolen on the way to the polls, or may not resemble the photo in the identification because he recently grew a beard.” Crawford, 553 U.S. at 197. The risk that a voter might be unable to cast his vote on account of this restriction did not constitute a severe burden. Similarly, nowhere did Crawford mandate that Indiana provide voters with notice and an opportunity to cure before they were turned away from the polls.[36] In fact, in Crawford the Court noted less burdensome methods of identification, including a requirement that voters “sign their names so their signatures can be compared with those on file.” Id. (emphasis added). The dissent lauded as “significantly less restrictive” a voter-ID system in which a Florida voter who lacks photo ID may cast, at the polling place, a provisional ballot that will be counted if the state “determines that his signature matches the one on his voter registration form.” Id. at 239 (emphasis added) (Breyer, J., dissenting). Nowhere did the dissent intimate that this “significantly less restrictive” voter-ID system required notice or an opportunity to cure before rejection. See id. By concluding that Texas’s signature-verification requirement does not constitute a severe burden—even without notice and an opportunity to cure—we join the Ninth Circuit, which agrees that “the absence of notice and an opportunity to rehabilitate rejected signatures imposes only a minimal burden on plaintiffs’ rights.” Lemons, 538 F.3d at 1104. This is so even where “county elections officials do not notify voters after rejecting nonmatching signatures.” Id. b. We next determine whether “the State’s important regulatory interests are . . . sufficient to justify the restrictions,” and they generally are, under Burdick, if the burden of the voting restriction is not severe. Burdick, 504 U.S. at 434 (cleaned up). We agree with the Secretary that Texas’s interest in preventing voter fraud justifies its signature-verification requirement. It is well established that the electoral process poses a risk of fraud. See Voting for Am., Inc. v. Steen, 732 F.3d 382, 394 (5th Cir. 2013). “[N]ot only is the risk of voter fraud real but . . . it could affect the outcome of a close election.” Crawford, 553 U.S. at 196. Thus, “[w]hile the most effective method of preventing election fraud may well be debatable, the propriety of doing so is perfectly clear.” Id. Texas “indisputably has a compelling interest in preserving the integrity of its election process.” Eu v. S.F. Cty. Democratic Cent. Comm., 489 U.S. 214, 231 (1989). But Texas’s signature-verification requirement is not designed to stymie voter fraud only in the abstract. It seeks to stop voter fraud where the problem is most acute—in the context of mail-in voting.[37] “[T]he potential and reality of fraud is much greater in the mail-in ballot context than with inperson voting.”[38] Texas’s important interest in reducing voter fraud—and specifically in stymying mail-in ballot fraud—easily justifies its use of signature verification. In concluding otherwise, the district court made at least two errors: It (1) incorrectly suggested that Texas needed to provide evidence of voter fraud and (2) erroneously imposed a narrow-tailoring requirement on the state. First, the district court deemed it relevant that “there is no evidence in the record demonstrating that any mismatched-signature ballots were submitted fraudulently.” Richardson, 2020 WL 5367216, at *34 n.44. But we do not force states to shoulder “the burden of demonstrating empirically the objective effects” of election laws. Munro v. Socialist Workers Party, 479 U.S. 189, 195 (1986). States may “respond to potential deficiencies in the electoral process with foresight rather than reactively.” Id. at 195–96. States have thus “never been required to justify [their] prophylactic measures to decrease occasions for vote fraud.” Tex. LULAC, 2020 WL 6023310, at *7. For instance, in Crawford, although “[t]he record contain[ed] no evidence of any such fraud actually occurring in Indiana at any time in its history,” the Court still concluded that “[t]here is no question about the legitimacy or importance of the State’s interest in counting only the votes of eligible voters.” Crawford, 553 U.S. at 194, 196. By intimating that Texas ought to provide the court with evidence of voter fraud, the district court ignored this court’s binding conclusion that “Texas need not show specific local evidence of fraud in order to justify preventive measures.” Steen, 732 F.3d at 394. Second, the district court misapplied the Anderson/Burdick methodology by erroneously imposing a narrow-tailoring requirement. Under Burdick, 504 U.S. at 434, where a voting restriction imposes a severe burden, the state must show (1) that there is “a state interest of compelling importance” and (2) that the regulation is “narrowly drawn to advance” that interest. But where the burden of an election law is reasonable—instead of severe—the state must show only a “legitimate interest[]” that is “ sufficient to outweigh the limited burden” imposed by the regulation. Id. at 440. The Anderson/Burdick framework does not impose a narrow-tailoring requirement on the state when dealing with reasonable burdens. Id. The Secretary satisfied her burden by proving that Texas’s interest in thwarting voter fraud justifies signature verification. The district court even suggested as much. Richardson, 2020 WL 5367216, at *35. But instead of accepting this important interest and weighing it against the burden on the plaintiffs, the district court imposed an additional burden: The Secretary must show that Texas’s interest in preventing voter fraud “is furthered by utilizing signature comparison procedures that do not provide voters with a meaningful opportunity to avoid disenfranchisement by curing an improperly rejected ballot.” Id. at *29. According to the court, Texas failed in this endeavor because there is “no rational basis for providing robust cure procedures to voters who fail to show an ID when voting in person but not those whose signatures are perceived to mismatch when voting by mail.” Id. at *35. The district court cited no authority for this added burden on the Secretary. And for good reason. In effect, the court required the Secretary to show that Texas could not have fashioned its regulations in a less burdensome manner. When we say that a state has not met its burden because we can imagine “less burdensome regulatory options [that] were available,” we call that a “narrow tailoring requirement.” McConnell v. Fed. Election Comm’n, 540 U.S. 93, 316 (2003) (Kennedy, J., concurring in the judgment). But the Anderson/Burdick framework imposes a narrow-tailoring requirement only on restrictions that constitute severe burdens, not on reasonable voting restrictions. Burdick, 504 U.S. at 434. The district court thus misapplied Anderson/Burdick when purporting to analyze reasonable restrictions on the right to vote.[39] Texas’s important interest in preventing voter fraud in its mail-in ballot system is sufficient to justify its reasonable restrictions on the right to vote. The Secretary is likely to prove that the district court erred in granting the plaintiffs’ summary judgment on the merits. B. The Secretary is likely to prevail in her defense that sovereign immunity bars the district court’s injunction requiring that she issue particular advisories and take specific potential enforcement action against noncomplying officials. Whether Ex parte Young, 209 U.S. 123 (1908), bars all affirmative injunctions against an officer “is an unsettled question that has roused significant debate.” Green Valley Special Util. Dist. v. City of Schertz, 969 F.3d 460, 472 n.21 (5th Cir. 2020) (en banc). We need not settle that debate here. Although the question remains whether sovereign immunity bars all affirmative injunctions, the present injunction is impermissible because it would control the Secretary in her exercise of discretionary functions. In Young, 209 U.S. at 158, the Court stated that “[t]here is no doubt that the court cannot control the exercise of the discretion of an officer.” Analyzing the question whether sovereign immunity bars positive injunctions against officers, the D.C. Circuit stated that “an attempt to control an officer” in the exercise of a discretionary function would violate sovereign immunity under Ex parte Young, and “would place the court on the wrong side of the line thought to divide ‘discretionary’ from ‘ministerial’ functions.” Vann v. Kempthorne, 534 F.3d 741, 753 (D.C. Cir. 2008) (citing Hagood v. Southern, 117 U.S. 52, 69 (1886)). The D.C. Circuit further examined the Supreme Court’s application of sovereign immunity in Hawaii v. Gordon, 373 U.S. 57 (1963) (per curiam). See Vann, 534 F.3d at 753. In Gordon, 373 U.S. at 58, the Court held that a court could not require the Director of the Bureau of the Budget to withdraw a report advising the federal government regarding which land the United States should retain under the Hawaii Statehood Act. Such an order violated sovereign immunity because it “would require the Director’s official affirmative action.” Id. The D.C. Circuit explained that Gordon exemplifies “the principle” that a court may not compel officers to take affirmative official actions that are discretionary. Vann, 534 F.3d at 753. We need not determine now whether affirmative injunctions are categorically barred by sovereign immunity. See Green Valley, 969 F.3d at 472 n.21. It is sufficient to note that, at the very minimum, a court may not “control [an officer] in the exercise of his discretion.” Young, 209 U.S. at 158. The district court’s sweeping order requires that the Secretary take several positive actions. In addition to requiring her to issue an advisory notifying local election officials of the district court’s constitutional judgment regarding the signature-mismatch laws, the order also gives the Secretary an ultimatum. It provides that she either must issue an advisory stipulating the detailed procedures that the district court imposed, or, alternatively, must promulgate an advisory requiring that local officials refrain at all from rejecting ballots based on mismatched signatures. Richardson, 2020 WL 5367216, at *38. Section 31.003 states that the Secretary “shall obtain and maintain uniformity in the application, operation, and interpretation of this code,” and that in doing so she “shall prepare detailed and comprehensive written directives and instructions” regarding Texas election laws. Because the statute uses the mandatory language of “shall,” § 31.003 imposes an affirmative duty on the Secretary to maintain uniformity regarding the application and interpretation of election laws. See Lightbourn v. Cty. of El Paso, 118 F.3d 421, 429 (5th Cir. 1997). “If a statute, regulation, or policy leaves it to . . . [an] agency to determine when and how to take action, the agency is not bound to act in a particular manner and the exercise of its authority is discretionary.” St. Tammany Par. ex rel. Davis v. FEMA, 556 F.3d 307, 323 (5th Cir. 2009). Though the Secretary has a duty to maintain uniformity, § 31.003 leaves her considerable discretion and latitude in how to do so. By prescribing detailed and specific procedures that the Secretary must include in her advisory, the district court impinges upon her discretionary authority in flat violation of Young. The fact that the district court’s mandated procedures were offered to the Secretary as one of two choices does not cure the order from infringing on her discretion. To the contrary, the very fact that the order gave her an ultimatum constitutes “an attempt to control the officer” and is, thus, plainly forbidden under Young. See Vann, 534 F.3d at 753. The injunction also stipulates that the Secretary must reprimand any local officials who violate the district court’s procedures and must “correct the offending conduct” per § 31.005. Richardson, 2020 WL 5367216, at *39. Again, the order far exceeds the limits of Young. A “[r]eview of the provisions of the Texas Election Code that refer to the Secretary’s role in elections reveals that most give discretion to the Secretary to take some action.” Lightbourn, 118 F.3d at 428–29. Interpreting § 31.005, we determined that the Secretary has considerable discretion under that provision. Id. at 429. Indeed, we observed that the law states that she “may take appropriate action to protect the voting rights of the citizens . . . from abuse . . . .” Id. (quoting § 31.005(a)). The district court directs the Secretary to take action against non-conforming election officials under § 31.005(b), a provision that specifies how the Secretary can enforce the Code against violations of voting rights. As in § 31.005(a), the language in § 31.005(b) is discretionary, stipulating that the Secretary “may order the person to correct the offending conduct.” (Emphasis added.) “Provisions merely authorizing the Secretary to take some action do not confer a legal duty on [her] to take the contemplated action.” Lightbourn, 118 F.3d at 429. Section 31.005 grants the Secretary discretion to take enforcement actions, and the district court cannot, therefore, compel such actions under Young. Thus, the Secretary is likely to prevail in her defense that the injunction is impermissible under Young.[40] IV. The other factors also counsel in favor of granting a stay pending appeal. As to whether the Secretary “will be irreparably injured absent a stay,” Nken, 556 U.S. at 426, “[w]hen a statute is enjoined, the State necessarily suffers the irreparable harm of denying the public interest in the enforcement of its laws,” Veasey, 870 F.3d at 391. And as to “where the public interest lies,” Nken, 556 U.S. at 426, when “the State is the appealing party, its interest and harm merge with that of the public,” Veasey, 870 F.3d at 391. Moreover, “a temporary stay here, while the court can consider argument on the merits, will minimize confusion among both voters and trained election officials—a goal patently within the public interest given the extremely fast-approaching election date.” TDP-I, 961 F.3d at 412 (cleaned up). Finally, as to “whether issuance of the stay will substantially injure the other parties interested in the proceeding,” Nken, 556 U.S. at 426, to whatever extent it might, it does not outweigh the other factors. “Our decision is limited to determining irreparable harm not in denying the plaintiffs’ requested relief outright but in temporarily staying the injunction pending a full appeal.” TDP-I, 961 F.3d at 412. Because of the likelihood that the Secretary will succeed on the merits, combined with the irreparable harm inflicted on the state and its citizens by the injunction, the balance of harms weighs in favor of the Secretary. * * * * * The Secretary’s motion to stay the injunction pending appeal is GRANTED. The injunction is STAYED in all its particulars pending further order of this court.[41] PATRICK E. HIGGINBOTHAM, Circuit Judge, concurring in the stay: In 1985, the Texas Legislature codified a revised state election code that included §§ 87.041(b) & (d), the provisions from which Plaintiffs seek relief.[42] Since codification, the Legislature has amended § 87.041(b) only once, in 1987.[43] Section 87.041(d) has not been amended. And while the Legislature has added to or amended other subsections of § 87.041 as recently as 2017, Texas’s basic framework for verifying voter signatures has been in place for several decades. Plaintiffs filed this suit in August 2019, reaching this court a year later. We are asked to change those rules while voting in a presidential election is under way—in the three weeks remaining before Election Day. However federal courts might finally decide this case, it now hangs a cloud over the election. I concur only in the decision to stay pending appeal of the district court’s injunction changing the election rules. The Secretary of State has shown a substantial likelihood of success on the merits, and the district court’s ruling has been stayed to allow this Court to decide the merits of the case. Well enough, but the reality is that the ultimate legality of the present system cannot be settled by the federal courts at this juncture when voting is already underway, and any opinion on a motions panel is essentially written in sand with no precedential value[44]—its reach is to delay, not to finally decide the validity of the state regulation. The Supreme Court has consistently counseled against court-imposed changes to “election rules on the eve of an election.”[45] Caution is particularly appropriate where, as here, the challenged laws were in effect long before suit was filed. I. I would not attempt to settle our circuit’s law on such complex and delicate questions in a preliminary ruling that has not benefitted from oral argument or collegial discussions. And a decision by this motions panel granting a stay settles no law. To the contrary, it has no precedential force and is not binding on the merits panel, leaving it as a writing in water—made the more empty by pretermitting the jurisdictional requisites of sovereignty and the reach of Ex parte Young. The matter is yet to travel its ordinary course to be settled by a fully considered opinion by the merits panel, perhaps then by the en banc Court. This reality is brought home by the changing opinions of my colleagues as the Court responds to legal challenges in the electoral process as conflicting opinions in other circuits indicate.[46] Here, we proceed without collegial conference on a motions panel and need not as a panel traverse numerous paths and crossroads engaging significant issues whose impact on our voting-rights jurisprudence remains contested, including standing and the reach of Ex parte Young, core principles of federalism. To do so would expose shifting views on these issues—a fluidity of view that unwittingly would present this Court as a volunteer in a political fight. In my view, the Secretary is a proper defendant under Ex parte Young. More to the point, it is the controlling law in this circuit. In pretermitting rather than accepting that reality, my colleagues cling to their view expressed last month that the Secretary lacks the enforcing authority under state law necessary to a federal suit enjoining her enforcement of an assertedly unconstitutional state statute[47] and casting doubt on whether the Court is bound by its recent case law because that case law might yet be considered en banc.[48] This fluidity counsels caution in wading into a change of election rules while voting is underway in an election charged with distrust of the political process—at its heart breeding doubt that one’s vote will count. II. In 2016 and 2018, “approximately 5,000 [Texas] ballots were rejected on the basis of perceived signature mismatches.”[49] Such “small” differences have the potential to decide both local and national elections. And with the large increase in votes cast by mail in our ongoing pandemic that error rate would toss out far greater numbers. There is much at stake here. While Chief Justice Marshall’s observation that the federal courts must decide is more than a truism, staying our hand is well within our compass here as we are asked to draw upon our injunctive powers. These must include an assessment of the real-world effect of when sought-for relief is granted. Plainly, the risks of now ordering changes in rules in effect for years would add to the uncertainties at every county seat across Texas, each facing the counting of votes cast by mail swelled by the pandemic beyond all past experience. There is yet another layer. A final decision from the judiciary is unlikely before voting in this presidential election year is completed. Again, it is now underway. Finally, while I cannot join Judge Smith’s opinion, I join the grant of a stay for the reasons I here offer. Relying on the old wisdom that looking to the path traveled can give direction to the road ahead, we see that while the road of right to vote has at times been nigh impassable as it rolled past people of color, women, and the poor, it has in the long view tracked the expansion of civil rights, reflecting to these eyes a maturation of individual liberty. Sometimes one step forward with two steps back, but the arc has been its expansion with which partisans ought make peace, accepting the bedrock principle that disenfranchising citizens is not a fallback to a failure to persuade. It is a given both that states must protect citizens’ fundamental right to vote, resisting in that effort tempting cover for partisan objectives, and that their efforts remain reviewable with the disinterest demanded by the architects of our Constitution, insisting that judges of federal courts it would create be as “independent as the lot of humanity will admit”[50]—counsel wise and prescient offered as it was before the arrival of political parties, a charge implicit in the oath of us all whether modern day federalists or Jeffersonians.
